采用希尔伯特振动分解的非整数次谐波检测新方法

Novel Method for Non-integer Harmonics Measurement Using Hilbert Vibration Decomposition

针对现有电力系统非整数次谐波分析法的不足,提出了一种改进的希尔伯特振动分解(HVD)方法。该方法根据对解析信号瞬时频率的分析,巧妙地通过平滑滤波获得非整数谐波成分中幅值最大分量的频率,由同步检测获得相应的幅值和初相角,通过迭代运算自适应地检测出非整数次谐波的各次频率、幅值和相角。虽然HVD方法和希尔伯特黄变换(HHT)方法这两者均以希尔伯特变换为基础,但HVD方法避免了复杂的经验模式分解(EMD)过程。采用Savitzky-Golay滤波替代平滑滤波,在保留有效频率成分情况下可极大地消除快速变化不对称振荡高频值;提出的新波形特征匹配边界延拓可消除边界效应的影响,使得非整数次谐波分析更准确。仿真实验证明了改进的HVD方法对非整数次谐波检测的有效性。

An improved Hilbert vibration decomposition （HVD} method was introduced to the non-integer harmonic analysis in electrical power system. The non-stationary frequency of the largest component was estimated as a smoothing filter of the instantaneous frequency of the composition, and the corresponding envelope was estimated according to synchronous demodulation. The frequency, amplitude and phase of every frequency component of noninteger harmonics could adaptively be extracted at each iteration step. The proposed HVD method was based on the Hilbert transform （HT） , just as Hilbert-Huang transform（ HHT）, but did not involve complicated empirical mode decomposition {EMD）. Savitzky-Golay{S-G） filter was applied to remove the rapidly varying asymmetrical oscillating part and to leave only the useful frequency. A novel wave characteristic matching extending method, dedicated to overcome the end effects of HVD method, was proposed to ensure the high precision of non-integer harmonics analysis. The validity of the improved HVD method was verified by simulation and experiment.